[ros-users] Interfacing a new arm driver to a motion planner

[Martin Günther]

On Sun, 5 Dec 2010 17:39:30 +0100
Martin Günther <mguenthe at uos.de> wrote:

> Dear list,
> 
> I am currently writing a ROS node for the Neuronics Katana 450 arm and
> would be happy about a few quick pointers about how to continue. I am
> aware of the ROS packages katana450, katana_ik_bridge and
> stair_katana; however, I would like to reuse as much ROS stuff for
> IK, arm navigation and motion planning as possible, so I decided to
> rewrite the code from scratch.
> 
> So far,
> - I have created a URDF description of the arm;
> - I can read out the current joint angles and publish them as
> JointState messages;
> - I can move the arm to a desired joint angle configuration;
> - I can move the arm to a target pose in space, using the inverse
>   kinematics library supplied with the arm (although I'm also eager to
>   give the ROS IK a try).
> 
> I have the feeling that I am almost done (right? come on, please!),
> but I'm a little overwhelmed by the wealth of arm navigation stuff in
> ROS. What are the messages/services/actions that I should provide to
> be able to interface my arm to a motion planner, and to use the ROS
> IK, and so on?
> 
> Best,
> Martin

Forgot to mention: The robot arm has realtime controllers for all the
motors; I can send a spline (or a sequence of splines) there, and it
will be executed.

Best,
Martin